Overview and details of the sessions of this conference. Please select a date or location to show only sessions at that day or location. Please select a single session for detailed view (with abstracts and downloads if available).
Please note that all times are shown in the time zone of the conference. The current conference time is: 27th Jan 2022, 09:32:53pm CET
Session Chair: Anne Feuer, University of Stuttgart, Germany
10:00am - 10:30am
Invited Talk: Applications of artificial intelligence and machine learning in laser materials processing
Laboratory for Advanced Materials Processing, EMPA, Switzerland
Laser materials processing (welding / additive manufacturing) are known to be highly dynamic. The reason is the non-linear nature of light-matter interactions. This not only complicates the reproducibility of the process quality in mass production but it is also a challenge for in situ and real-time quality monitoring and control.
Under such circumstances, our approach has been to record signals from different sensors such as acoustic emission and optical sensors. However, due to the complex character of the signals, traditional signal processing methods are limited to extract the useful information about the process quality. To overcome this difficulty, we use state-of-the-art artificial intelligence and machine learning methods as they allow building complex empirical models from complex structured datasets.
The presentation makes an overview of our approach and results for laser processes.
10:30am - 10:45am
Determination of the beam position in laser deep penetration welding using coaxially acquired images of the keyhole front and machine learning
Pablo Dilger1,2, Carola Forster1, Elias Klein1, Silvana Burger1,2, Eric Eschner1,2, Michael Schmidt1,2
1Institute of Photonic Technologies (LPT), Friedrich-Alexander-Universität Erlangen-Nürnberg, Konrad-Zuse-Straße 3/5, 91052 Erlangen, Germany; 2Erlangen Graduate School in Advanced Optical Technologies (SAOT), Friedrich-Alexander-Universität Erlangen-Nürnberg, Paul-Gordan-Straße 6, 91052 Erlangen, Germany
The joining technology of laser beam welding offers high flexibility and productivity. However, the small laser beam focus demands dependable quality assurance to ensure a sufficient connection of the parts. In keyhole welding of metal sheets in butt joint configuration, a gap is visible at the keyhole front, which correlates with the leading joint position. This process feature can be used for quality control by arranging a high-speed camera coaxially to the laser beam to monitor the keyhole. Here, we present a machine learning approach for a robust determination of the beam position relative to the joint based on the keyhole front morphology. For this purpose, we conducted a series of experiments to produce a set of labeled images, which are used to train a convolutional neural network. After training on the data the network can predict the keyhole front gap position, setting the foundation for a quality control system.
10:45am - 11:00am
Use of hyperspectral imaging (HSI) in combination with machine learning methods for the critical powder parameters and corresponding part properties. prediction of
Martin Schäfer1, Christoph Wilsnack2, Florian Gruber2, Axel Marquardt3, Sebastian Witte2
Additive manufacturing processes are generally operated in an industrial environment with defined parameters for specific materials and specific applications. The machines used can only be individually controlled to a limited extent. This means that the material properties are directly related to the component qualities. The evaluation of materials before and during the construction process is thus an essential component in the quality management of the AM production chain and the basis for optimized use and reuse of production value materials.
The analysis of metal powders by HSI represents a potential novel method for the qualification of powders.
To this end, the use of hyperspectral imaging, in combination with machine learning methods, for the prediction of critical powder parameters such as the rheology, the morphology and the chemical properties of the powders will be demonstrated. Furthermore, it is discussed if a prediction of the component properties by HSI sensing.